When people think of debt collection they think of aggressive and often relentless phone calls and mailed notices. They don’t think of sophisticated machine learning techniques using millions of data points to figure out the best collection approach. Our next guest on the Lend Academy Podcast is out to change that.
Ohad Samet is the CEO and Co-founder of TrueAccord. They are a new kind of debt collection company that uses a data driven approach and digital first communications. Rather than a large team of telephone operators TrueAccord has found it is far more effective to let technology do the hard work.
In this podcast you will learn:
- The call from a rude debt collector that led to the founding of TrueAccord.
- The state of the debt collection industry today and how it is changing.
- How the traditional debt collection process works.
- How TrueAccord is bringing innovation to this process.
- How their system uses machine learning.
- The way they make money.
- How much better their system is than traditional agencies.
- The verticals where TrueAccord is focused.
- The scale they are at today.
- Why compliance is the most important feature of their product.
- How they are able to reduce compliance risk.
- Why Ohad decided to join the CFPB Consumer Advisory Board.
- What the fundraising process has been like for TrueAccord.
- What the future holds for TrueAccord.
This episode of the Lend Academy Podcast is sponsored by LendItFintech USA 2018, the world’s leading event in financial services innovation.
Download a PDF of the transcription of Podcast 135 – Ohad Samet.
[expand title=”Click to Read Podcast Transcription (Full Text Version) Below”]
PODCAST TRANSCRIPTION SESSION NO. 135-OHAD SAMET
Welcome to the Lend Academy Podcast, Episode No. 135. This is your host, Peter Renton, Founder of Lend Academy and Co-Founder of LendIt.
(music)
Today’s episode is sponsored by LendIt USA 2018, the world’s leading event in financial services innovation. It’s going to be happening April 9th through 11th, 2018 at Moscone West in San Francisco. We’re going to be covering blockchain, digital banking and of course, online lending, as well as other areas of fintech. There will be over 5,000 attendees, over 250 sponsors and registration is now open. Just go to lendit.com/usa to register.
Peter Renton: Today on the show, I am delighted to welcome Ohad Samet, he is the CEO and Co-Founder of TrueAccord. They’re an interesting company, TrueAccord, they’re in the debt collection space, but they are very different to the traditional debt collectors that we all know. They have a different approach that’s steeped in technology and machine learning and we go into that in some depth. They also don’t consider themselves real debt collectors, they’re more about developing a relationship with the customer.
It’s a unique approach that I think in all of the time I’ve been in this industry, I haven’t seen anyone’s approach quite like TrueAccord. So we dig into that, we talk about how they actually go about communicating with the customer, what they do differently, we talk about compliance because this is an area that really…you have to have all of your ducks in a row where that’s concerned. We talk about the different target markets that they’re going after and of course, we talk about where Ohad is going to be taking this company. It was a fascinating interview, I hope you enjoy the show.
Welcome to the podcast, Ohad.
Ohad Samet: Thank you for having me, really glad to do this.
Peter: Okay, so let’s just get started by giving the listeners a little bit of background about yourself. It sounds like you‘ve been in fintech since before the term fintech existed and so why don’t you just tell us what you’ve done before you actually started TrueAccord.
Ohad: Yeah, sure. I’ve been in machine learning for financial services for almost 13 years now, in fact. Started as one of the early employees in a company called Fraud Sciences. We did machine learning for fraud prevention and e-commerce before it was a big deal, that was in 2005. I was their Head of Analytics and in 2008, we got acquired by PayPal so I had the opportunity to work for PayPal for a few years. And then in 2010, I left PayPal and started working on two projects that became companies.
One of them is Signifyd, one of the leaders in machine learning for fraud prevention in e-commerce now, kind of a Fraud Sciences 2.0, you know, I’ve been part of the early days and the team has taken it so far, so much further than I would have, but I was there in the first few days, first few months.
The other one was called Analyzd, we basically were involved in machine learning for consumer credit underwriting and again, a little bit before lending became a big deal and that one got acquired by Klarna, a European payments company in 2011. I was Klarna’s Chief Risk Officer for almost two years, we were doing about $2.5 billion a year in payment volume, buy now, pay later and that was a great experience, a great hyper growth experience, I learned a lot there. In 2013, early 2013, I left Klarna and started working on a few projects and the one that matured was TrueAccord, that we raised money for in 2013 and launched in 2014.
Peter: Okay, so then let’s just go back to that time period and wanted to sort of…obviously, you were Chief Risk Officer at Klarna, I mean, we know Klarna pretty well. In Europe, they’re very, very big where they have online payment plans for online purchases and was that where you first saw this kind of behavior that felt like led to TrueAccord, as far as the fact that debt collection could be done a lot better? What was the seed that started TrueAccord?
Ohad: Yeah, I always tell the story that I got a call from a rude debt collector which is true. I owed about $150 on a retail credit card that I forgot to pay and I got a call and it’s not that that person was a criminal or anything, but it was just not a good experience and I started thinking about what it was like for people who can’t afford to pay, who are hounded by five or six different debt collectors, but it was also the business experience, absolutely.
Looking to what we did at Klarna, we were so sophisticated in customer acquisition, so sophisticated in underwriting and in collections…not to say that the people there haven’t been or are not sophisticated now, but there was…we still stuck to the call center model. I looked around at other financial institutions and even upstarts in the lending business and everybody seemed to have stuck to the same call center and letter model for debt collection and it didn’t make sense. It felt like a type of problem that I knew how to solve in terms of the machine learning approach. It’s very similar in my perspective to fraud prevention and credit underwriting.
It felt like an opportunity to do some social good because the formal numbers are 77 million people a year in the US are subjected to some type of debt collection activity. That’s a lot of people and that’s not a great experience. It also felt like a market that was making a lot of money, you know, $15 billion a year, more or less in revenue, highly fragmented with a bunch of mom and pop shops and not a lot of introduction of technology. It felt like it was looking for innovation, it was looking for new technology and nobody really wanted to touch on it because of how regulated it was.
So as an early team, we felt like one, we understood the machine learning and the product aspect that needs to be brought to this market. We saw a huge business opportunity, we weren’t afraid of diving deep into the regulatory aspect and the compliance aspect of the business. We felt very well positioned to do something meaningful in the debt collection market and turn it from a very transactional…I’m going to call you as much as I can until you pick up and get you to call to pay to a more relationship driven approach that’s email first, digital first, it’s personalized, it responds to consumer preference and behavior and that’s what we did.
Peter: Okay, so then would you say like if you look at the debt collection industry today, it’s still pretty much the same as it was 10/20 years ago. Do you feel like…has there been, I mean, you’ve now been doing this business four years or so, do you feel like there has been a shift in the last four years or is it still pretty much the same as it always has been?
Ohad: There’s definitely a shift, but the shift is not in introducing new technology. The shift is in how the industry responded to the CFPB, to new rules, how the industry, at least the parts in it that works with credit card issuers, responded to additional requirements from banks based on the OCC’s requirements for vendor management, there are trends in regulation and case law around debt collection.
But in terms of people coming up and saying, hey, we’re going to reinvent this market, we are going to step away from the old model, not really, no. We are sitting in a really unique intersection. On one hand, we’re really excited about the product and about the technology aspect of the business; on the other hand, we are really willing to dive deep into the regulatory compliance aspect and listen, it’s very difficult to do. Like replicating what we did is going to take several years, a lot of money and a lot of effort so I understand why there hasn’t been a lot of movement in that market.
Peter: Okay, so then I’d like to…if you could maybe walk us through your approach and how it’s different, how you incorporate machine learning. I mean, we all know it’s a pretty simple business in many ways. As you say, people phone you up or they send you a letter and so how is your approach different?
Ohad: Let’s start with how it works with a traditional collection agency. If you are sent to a traditional collection agency, the first thing you’re going to get probably is a letter that hopefully reaches you, snail mail is what we call it in the industry, right, in the technology industry, hopefully, it reaches you. It has all the disclosures and everything that you need to know about your debt and everything that compliance requires. After that, you start getting phone calls and in some instances we’re talking about four to six call attempts per day and the goal of those phone calls is to get you to respond and to get you on a call with an agent which is basically a cold sales call. They get on the phone and they do their best to get you to make a payment for the debt that you owe.
The payment options that they offer you are governed by the constraints they get from the creditor and are also governed by the fact that the agent that you’re talking to is commission based. They have a very low base and if their collection goes then they get a higher salary. As a result, the experience, even when it’s the most compliant, even when the people on the phone are highly experienced, it’s not a great experience.
It’s an experience where it’s not uncommon for collectors at the end of the month to be more pushy and more aggressive, even if they do that in the most compliant way. It’s an experience where you get calls like I said several times per day, you get phone calls from numbers you don’t recognize and they happen sometimes during not convenient times. It requires you to get on the phone with a human which is usually or very often embarrassing to discuss your debt situation, people don’t like that, don’t appreciate being on the phone with another person. It requires a bunch of stuff.
Now in comparison, you could also…so this is kind of like a telemarketing campaign when you think about that. A collection company is like a telemarketing campaign; call a lot, get people on the phone, get them to the sale.
In our case, it’s a lot more like a really sophisticated inside sales department. When we get an account, we have an engine that’s called “HeartBeat,” that’s our product. HeartBeat looks at you as a consumer and compares you to one of the millions of consumers we’ve had in our platform, so far, looks at features that we calculate for you, for your debt, for everything we know about you from the creditor. We don’t collect social data information or anything like that, we just use the information we get from the creditor.
Based on that, it says well what is the best way to contact this customer; what day, what time, what channel, what content to put in front of them and then it starts with an initial communication. Usually, the first, first communication is an email because we need to get again all of these disclosures in to make sure that we are completely compliant and then based on your behavior, it will decide what the next step needs to be. It tracks everything that you do in response to our communication, every email you open; every link you click on a text message or an email or a push notification, your browsing pattern on our website, which pages you spent time on, which offers you looked at, did you investigate certain payment options versus others.
Did you call in and talk to an agent, what was your disposition, what was your emotional state and it crunches all of that in real time to decide, how do I follow up with this consumer. If they clicked in an the email, looked at a couple of payment plans and then dropped off, maybe I can offer them a more flexible payment plan. If they called in and they said that they’re in financial distress and they can’t pay for the next few weeks, well, I don’t need to email them or try to contact them for a few weeks. When I come back, I should start with content that talks about their context and what’s happening in their lives so it taps into hundreds of contact items across multiple channels to personalize the experience and talk to the consumer in context.
Also because it’s highly automated and scalable, it can offer highly customized payment plans. So if a consumer wants to pay because they have two/three jobs and their paycheck is not uniform across weeks or across months…they want to pay $50 this month and $75 next month and then $30 and then $100, we can do that because our system will administer that in a predictable and scalable way so there is this personalized and highly customizable experience. At the end of the day if you do call in or in the rare case that we called you and we get you on the line, you talk to a team that’s very different in terms of the people we source and the training we give them.
They absolutely get all the the compliance training, but they don’t get commission. Their focus is on customer service and walking you through the process and helping you work with our system so overall, you get treated like a customer and that shows in the way consumers respond to us. We do consumer interviews and people start thinking of themselves as our customers. Which is great because it’s conducive to our overall mission which is we don’t only want to help you with this debt, we want to help you with your overall financial life and how to progress yourself, how to manage your cash flow, how to think about your financial decisions so that you don’t end up in debt again, or if you do, you reduce your debt liability.
Peter: Okay, that makes sense. So where in all of this…is it the mere fact, you talked about…obviously you’re tracking everybody and you see there are millions of people that you can compare these customers to…I mean, is this where the machine learning specifically comes in or can you explain where that component kind of makes a difference in what you’re doing?
Ohad: The machine learning component?
Peter: Yes, machine learning component specifically.
Ohad: Well, what the system tries to do is it tries to replicate the way a human collector would respond to what the consumer is doing. So when I say that the system decides what the next step should be, what day or what time should we email a person or should we message a person, what channel should we use, what content should we use from the hundreds of content items we have, this is all done not by rules that are put into the system, not by manual decision, it’s done by an actual set of algorithms, an actual ensemble of algorithms that use historical data and consumer behavioral data to decide what the next step should be.
So when it looks at a specific consumer, it takes in all of the consumer’s previous actions, all of the information we have about the debt and about the consumer. It uses a feature vector to calculate in-house, it uses clustering algorithms to find similar consumers and calculate what is the path, what is the action it’s most likely…highest probability to get them to actually move forward towards a commitment to pay for their debt and then it chooses that action.
So it’s all data driven, automated and doesn’t involve people writing in rules to say if this, then that and that’s where machine learning plays into it. Having done this for fraud prevention, having done this for credit underwriting, I knew and our numbers and our results prove that at scale, machines are much better at deciding what the next steps should be based on historical data, what the treatment should be than humans.
Humans are good at edge cases, but on a day-to-day basis humans get tired, they get angry, they get distracted; we suffer multiple biases because we’re human, a machine doesn’t. If it needs to crunch 2 million interactions or 100 million interactions to decide what the next step should be and make the best decisions based on data without being distracted or being angry or being emotional, it will do that. And so machine learning is at the heart of our system because we want to deliver personalized experiences, consistent and personalized experiences at scale to consumers and that’s how we do it.
Peter: Right, that makes sense. So I wanted to just get a sense…I would like know, are you a traditional collection agency insofar as you are taking ownership of the debt? How does your model work because I know you’re talk here about owning the customer so does that mean you’re buying the debt or what’s the relationship you have with your clients?
Ohad: Yeah, we don’t really own the customer, but we create a trusted relationship with the customer.
Peter: Right.
Ohad: We don’t buy the debt. In fact, it’s a common misconception. There are debt buyers and there are debt collectors. Companies that buy the debt sometimes do their own collections, but many times just turn around and give it to a servicer, to a debt collection servicer and that’s what we are. We service the debt, we don’t buy it.
If a company like say Yelp or, you know, we don’t have a lot of clients that are public about using us, but some of the major banks, of course, are using us. If they need help with debt collection, if they use a third party or a first party debt collection company which means someone that collects under their own brand or someone that collects under the bank’s brand, we’re one of those options. We’re a fully licensed collection agency, we collect like a collection agency, we compete head-to-head against traditional players and we win and we get more business.
Peter: Yeah, so on that, do you have any examples you can share? I imagine this is quite measurable because…I mean, you can certainly measure the dollars that have come in, you can probably also measure the satisfaction of the people who are being chased by TrueAccord as opposed to a traditional collection agency so do you have any examples of stories you can say…oh, we went in here and…I imagine companies don’t just give you their entire book, they will say, here is 10% of our loan book, of our collections book, go and work on that and see what happens. Can you tell us sort of how you’ve done those kinds of head-to-head tests?
Ohad; Yes, so it’s hard to name specific clients. Debt collection is still a very sensitive area, but usually we come in, we get say 5% to 10% of the volume and we are measured at the end of the day, at the end of the placement period because we don’t get that relationship with the customer for a very long time. It can be 90 days, it can be 180 days. At the end of the period they look at us versus our competition and they look at how much we’ve collected, what percentage of dollars that we collect out of the dollars that they gave us versus the competition and usually, we are 50% better and at times in edge cases we could be as high as 5 times better than the competition. And that is, again, because consumers choose to pay with us because of the treatment and because of the flexible payment options.
Peter: Right, so you mentioned Yelp, you’ve mentioned banks, I mean, if you’re on your website you also have LendUp and we’ve had Sasha on the show here before. What verticals are you most focused on?
Ohad: We are focused on credit cards and consumer lending so unsecured consumer lending. We’re probably going to expand to additional markets this year just based on the traction that we’re seeing, but that’s where we’ve had our traction historically,
Peter: Okay, can you give us some sense of the scale you’re at today, I mean, how many people? I know you probably are not going to share with me revenue numbers, but just give me some sense of the scale.
Ohad: Yeah, so we are about 80 people, we have more than 2 million consumers that have passed through our platform, almost $2 billion of debt that we’ve been working on with about 80 clients ranging from very large banks and credit card issuers to quite small e-commerce shops. We’re a growth stage company so post Series B, whatever that means in terms of revenue and growth.
Peter: Right, right. Okay, you’ve talked about this a bit and I know that we’ve got like a Fair Debt Collections Act or whatever it is…compliance is, you’ve got a cross your T’s and dot your I’s, it’s really important to not get that wrong. Tell us about your approach to compliance, is this something you’ve got to approach in sort of brute force fashion where you just got to have all the different pieces in place or can you apply technology to make it more efficient, I mean, how are you approaching it?
Ohad: There is no cutting corners in compliance.
Peter: (laughs) Right.
Ohad: Not at all. We are regulated on a state level, on a federal level. Our clients are very demanding in their audits, in their compliance oversight which they should be so no cutting corners. We have a quite large…for the size of company that we are, legal and audit department, we have multiple outside counsel, we have a robust compliance program, there’s just no cutting corners. This is no joke, this is one of the most important features of our product, the product needs to be compliant, otherwise, nobody cares about returns and user experience. That’s the most important thing and we’re committed to that.
In terms of introducing technology, well, we’re more open to using vendors to manage compliance risk, I guess, than building things in house. We can use the service in AWS instead of having them in a server rack in our office which is still the standard for a lot of folks in the industry. I can say that because it’s called control compliance, because of the way our system is handled, it’s easier to reduce compliance risk because all of our content is pre-written.
It’s easier to approve it in advance and say I’m comfortable with this content, I’m comfortable with what we’re telling the consumer instead of having millions of calls a month and hoping that nobody says anything wrong on the phone so that’s one.
The second thing is if there is a change in regulation or some specific quirk that a client needs, they have some requirement in some state or what have you, it’s much easier to code something into the system than to try and retrain a hundred or a thousand agents to remember that when they talk to consumers on the phone. In that regard having the technology infrastructure is very helpful.
Peter: So does that mean you don’t have people who call, you’re doing everything electronically?
Ohad: No, we do have a small call center.
Peter: You do, okay.
Ohad: Yeah, but it is really tiny, it’s less than 20 people whereas if we were a traditional collection agency we would have hundreds and hundreds of people. The majority of what they handle is inbound communication, people who call in and want to talk to someone so it’s more of a customer care angle. If we call someone we try to call once or twice a month so it’s really, really infrequent that we call.
Peter: Right, right. Okay, so I want to switch gears a little bit and talk about the CFPB. You recently joined their Consumer Advisory Board?
Ohad: Yeah.
Peter: Can you tell us a little bit about what that does and why you joined it?
Ohad: Well you know with the caveat of leadership changes in the CFPB, I don’t really know… (Peter laughs)
Peter: Of course.
Ohad: As uninformed as any of the listeners is about that. Look, I believe in engaging with your regulator early and often. We have been in touch with the CFPB since 2013, we believe in our mission, we believe that everything we do is kosher, that we don’t need any carve outs to do what we do. So it was an easy decision to engage with the regulator early on especially because from 2013 there was an active rule making process for the debt collection market and we wanted to make sure that technology is well represented.
For me, getting on the Consumer Advisory Board is a continuation of that. One, debt collection needs to be represented; two, the technology sector needs to be represented; three, I’ve been in, again, financial services for more than a decade so obviously I have strong opinions about a myriad of topics and issues in financial services in the US.
Also, that fits into our larger goal at TrueAccord where we want to really help people build the financial lives that they deserve and that means debt collection is just an entry point. That means so much more, that means thinking about the underbanked and unbanked, that means thinking about females and how people don’t have access to traditional financial services, thinking about non-profits and how they fit into the picture, thinking about housing situations and how they fit into the picture.
All of that makes membership in the Consumer Advisory Board a really interesting place to be able to impact thinking as much as possible. People on the Consumer Advisory Board have no formal decision process, we only advise, we’re exposed to some things that are not public but only for a short period of time so you can’t say that we have more power than anyone else, we don’t, but at least we’re in the room to have the discussion and I think that’s very important.
Peter: Yeah, that makes sense. I wanted to just talk briefly about your Series B last year, late last year. I think it was a $22 million Series B that you closed and wanted you to talk us through the fund raising process. Was it more difficult than in previous times or were you finding your ideas getting an easier reception?
Ohad: It’s very binary with investors because it’s a polarizing market and a polarizing idea. When you look at our company…if your takeaway is, well, these guys are reinventing something, it’s really broken, this is why I’m in technology to reinvent these issues, to solve real problems for people and I understand that this is a big market then you get really excited and getting you onboard is easy.
I think Arbor Ventures, specifically Melissa and also Wei, they get that, our current investors get that. That’s why with every new round our current investors continue to double down on the company. In that sense, it’s pretty straightforward; on the other hand, there is a…if you look at some of the VCs that have looked into the company I have had with every round, with every round I’ve had similar conversations with people who say, we love you, we love the company, we love the idea, can’t invest in debt collection, we don’t know how to think about it and I understand that.
I respectfully disagree, I’ve been told that before; a few years ago, somebody said, we like to invest in brands that people like, like Uber and you know, I look at what Uber represents now in terms of the company’s reputation and I’m not sure it’s a loved brand to the extent that it was a few years ago.
So I understand where people are coming from, I understand that it’s hard to invest in a market that’s not hot right now and I absolutely respect that. I’m happy that there are investors that get what we’re trying to do, that get behind the mission, that are excited about the business so it’s been pretty straightforward for us to raise money and I’m happy for that and I’m very thankful.
Peter: Okay, so we’re almost out of time and I want to talk a little about the future here. You mentioned a few minutes ago that debt collection was really an entry point for your business and there’s lots of other things that you want to be doing so can you just take us through like what’s next…is there a new product on your roadmap or where do you go from here?
Ohad: So the way we think about our mission is now we have 2 million consumers on our platform. In several years, because we are expanding at the rate that we’re expanding, maybe we have 10/20 million consumers on our platform. They come back because there’s a level of recidivism in debt and we will see them multiple times. We ask ourselves, when we have seen almost everyone, when we have seen tens of millions of consumers, what do we want to be for them?
The answer is that we don’t want to just be the friendliest debt collector around or the most flexible debt collector around, we want to actually support them in their journey to get to the financial lives that they aspire to, we want to help them build equity. So if we get 40,000 people who tell us, I would pay you if I had a job, then maybe there’s a way for us to find a job for these people so we are working on an experiment with a partner to help people work from home and pay for their debt that way and that’s actually working very well. If people are saying, I have a housing situation, I’m homeless, or I have a problem I can’t make rent then maybe there is a non-profit in their area that can help them with housing so we’re thinking about that and so on and so forth even to the very basic.
When we interview consumers they say, well, thank you for the debt collection process, I paid my debt, can you help me rebuild my credit score, can you help me think about my cash flow then maybe it’s not us that do that predominantly, but we give access to other really exciting startups that are out there that do that type of work and we become a platform to give them access, but we have a window into an aspect of the US economy and some points international economy that not many companies know how to address.
We have a business model because at the end of the day we’re paid by their creditors. Their creditors turn around and every dollar that we collect they give us a cut of that so we’re actually not taking a cent from the consumer. Whatever the consumer pays goes to pay for their debt. We’re in a great position where our interests are, at least to some extent, highly aligned with the consumer, to help the consumer be in a better financial spot and that’s our long term goal, thinking about ourselves and that platform that has unique relationships with consumers who want to build equity, who want to build their financial lives and enabling that.
Peter: Fascinating, well, I certainly wish you all the best, unfortunately, we’re out of time. I really appreciate you coming on the show today, Ohad.
Ohad: Thank you very much for having me, this has been great.
Peter: See you.
Ohad: See you.
Peter: It really is fascinating to me that an industry as old as debt collection and let’s face it, debt collection has been around for thousands of years since the first company decided to give out credit to somebody there have been debt collectors. Certainly, the industry itself has had a bit of a bad rap and I think justifiably so with some of their tactics that is why it’s heavily regulated today. It’s just so refreshing to me to hear an approach that has really taken a friendly approach, but also a highly technological approach to make it a better experience for the customer. That’s really what TrueAccord is doing here, they’re trying to make it so you’re not feeling like someone’s hounding you, but you realize you have an obligation and they’re trying to make this a positive experience rather than a negative one.
Anyway on that note, I will sign off. I very much appreciate your listening and I will catch you next time. Bye.
Today’s episode was sponsored by LendIt USA 2018, the world’s leading event in financial services innovation. It’s happening April 9th through 11th 2018 at the Moscone West in San Francisco. It’s going to be the largest ever fintech event held in the Bay Area with over 5,000 attendees expected. We’ll be covering online lending, blockchain, digital banking and much more. You can find out more by going to lendit.com/usa.[/expand]
You can subscribe to the Lend Academy Podcast via iTunes or Stitcher. To listen to this podcast episode there is an audio player directly below or you can download the MP3 file here.
